THE BOOK--Playing The Percentages In Baseball

Another approach, which we followed in a league that I ran several years ago, was as follows: 

The first two rounds were snake-ordered.  Then, the third round picked in the same order as the second round.  After that, it resumed snake-ordering.

It seemed to work pretty well.

I’ve always wondered about this, especially in Football.  I know you don’t get into football as much on this site, but I suspect that Football has an even steeper slope (or at least did before the widespread implementation of 2-back systems).  Is there any way you could do a similar analysis on a 12-team football snake draft?  What about before and after the 2-back system became popular?  Any ideas on this?

I’m limited to the data at hand, and what I have is 22 forecasting systems in baseball.

Wouldn’t the discrepancy be eliminated in the forecaster’s challenge since you had 10,000 different drafts/leagues?

...

Milsy/#2--click my name for an article about football draft slots.

It looks like the slope on Tango’s graph is about 3 points/spot, whereas the numbers in the article above show that the slope is about 4.8 points per spot. (Although, the way the points are calculated, the slopes are about the same if you make the point scales equal to each other.)

Zach: yes, the idea of having 1000 drafts would be to make sure everyone drafted 40-50 times in each slot, and made sure you didn’t always follow the same participant.

I was just pointing out that had I not done that, then there’d be a severe bias in terms of figuring out who was better, since knowing where they drafted would be a big part of it.  And of course, in typical leagues, there is only one draft, not 1000.  So, every draft league has a bias to contend with.

Tango, perhaps the issue is simply too many teams in the draft?  The difference from #1 to #22 is LARGE.  Now if you have an 8 person draft, how bad is the #8 position compared to the #1?

Zach/4,

Thanks very much for the article.

to be fair, most fantasy drafts aren’t 22 teams deep.  I would assume it’s a lot less dramatic for a 12 team league.

Anyone else have trouble with Zach’s link?

I actually did, but I went to ESPN.com and got into similar articles.  Perhaps you need to be an “Insider” for it?

I don’t find the conclusion surprising at all . . . almost any attempt to assign value to picks (e.g. the NFL’s draft pick trade value chart or any fantasy auction) shows a rapid decline in value for picks after the first with the decline flattening quickly a few picks in. 

I am surprised you got such a nice curve with this season’s data considering how many injuries / disappointments there were among the top predicted performers.

Remember, the pick list is based on the order they were actually picked, not on the consensus order.  So, Pujols appears a bit through each of slots 1 through 6.  Dice-K, Wang, etc, likely appear pretty evenly throughout each slot.

I would think that it depends on the distribution of value from the players in the draft pool.

For example, if you have one player who is by far and away the most valuable fantasy player, then of course it pays to have the first overall pick in the draft.  Then let’s say you have three players who are of equal value and are considered to be the most valuable.  There would be no advantage to picking first in the first (and most important) round, vs picking third.  The person picking third then gets to make his 2nd round selection before the person picking first.  Once you get into the later rounds the difference in value between picks is smaller.

Another factor that should be taken into consideration is the knowledge of the other members of the draft.  If I can draft 6th and still get the 3rd best player, or draft 10th and still get the 5th best player, then that changes things.

Having the first overall pick is nice in many cases, but there are no “overlays” with the first overall pick.  If you pick Nth in the first round and N=1, then the Nth best player is the best you can do.  If you pick N=9, you can still conceivably (depending on the knowledge of the other league members) get the 1st to Nth best player, and pick higher in the second round before the player values merge closer together.

vr, Xei

Xei: I am talking about specifically the guys entering MLB 2009.  It looks to me like a pretty stable distribution, that there’s no outliers or whatnot.

And, there is no knowledge of other members in my draft either.  I don’t know why people are talking about this as a possibility.  Maybe it exists in other leagues (how?  I don’t know), but it’s not the case here.

So, my graph is specific to my forecasters, though I would not be surprised if it’s true generally speaking.

#14. Ok, the headline of the blog entry is a little confusing then.  Because in a “fantasy baseball snake draft”, there are other things to take into account.
vr, Xei

Dude, c’mon.

The headline says this:
“Is the fantasy baseball snake-draft fair? “

How long should I make my headline with how many conditions, especially since the very first line in the thread says:

“No, not based on the results of the Forecasters Challenge.”

If you want to suggest that it is especially unfair when there is a gap in talent, then fine.  But, given the data at hand, with the results we see, the draft is unfair.

Now, you also want to suggest “knowledge of others’ lists”.  How the heck does that happen?  I will further presume that even those casting their picks don’t even know who they will pick, that they don’t even follow their own scripts.

I’m annoyed because it seems that the data presented is simply being ignored in favor of theories that has no evidence presented.

Please, get back to the thrust of the thread.

For a fantasy league using a snake draft, I’ve always thought a good solution would be for teams to be able to bid for slots, using later picks as currency. Whichever team is willing to give up the highest later pick gets slot #1.

example - Slot #1 comes up for bidding, player1 bids his 15th round pick, player2 bids his 14th round pick, ... player1 bids his 7th round pick. No other bids.
So, player1 would get the 1st pick in the draft and draft from slot 1 in the normal snake manner, but would be skipped altogether in round 7.

That way, whoever values Pujols (or whoever else is the consensus #1) the most, gets the first pick.

cann: I like that idea.  And of course, no reason to limit this to only the 1st pick. 

You could have a general function where say the 1st slot loses his 7th pick, the 2nd slot loses his 8th, and so on.

I am talking about “real humans” drafting.  I am not talking about just a “points based” draft, which most fantasy baseball leagues are not.

I got a “Dude C’Mon!” and a “ridiculous” out of both blog authors in the same week.

I think the people who have played “fantasy baseball” for 20+ years realize that snake drafts are not fair, but to run your little “forecasters challenge” and present results without taking into consideration “human factors” in an “actual” fantasy baseball draft, with “real people” making the picks, which is how most drafts take place and that there are different league configurations available than a “points” league… your blanket statement may not be correct in some moderate cases.

Sorry if you don’t like the criticism.  At worst case, we may be talking apples to oranges, and that’s why I said the headline (and even the first sentence) was a bit misleading to us fantasy baseball players.

Yes, for your little challenge and computer automated draft scripts, the snake draft is not fair and I believe your data on that.  Beyond that, I am not convinced, nor have you proven anything.

vr, Xei

I think garik #6 makes a good point.  I’m assuming that this effect results from the dropoff in the first (or early) rounds being greater than the dropoff in subsequent rounds.  That would suggest that the effect would be lessened if you had a smaller league. 

Of course, looking at the graph, it appears that the dropoff in the second half of round one is even greater than the first half, which seems weird.  It makes me wonder if drafting toward the end of an 11 team league would have been more beneficial than drafting first (at least this year).

We are talking apples and oranges, and let’s leave it at that.

For the number of teams in the league, I find it fine that leagues have 8-12 teams in an AL-only or NL-only league.  But, it is beyond silly to have a 12-team league based on 30 MLB teams.

So, the 22-team league I have for MLB would be similar to my having an 11-team league for AL-only and NL-only.  Therefore, I don’t see the number of teams as having an impact.